Skip to content

Voice of Customer Themes Digest

Example prompt: "Every Monday morning, pull last week's resolved support tickets and group them into themes. For each theme, give me the count, a couple of representative customer quotes, and a one-line summary. Post the digest to #product-feedback on Slack and save the full version to our Notion product wiki."

How to automate voice of customer reporting with GloriaMundo

The Problem

Support tickets are the most honest product feedback we get, and they go almost entirely unread by the people who could act on them. Volume and SLA reports tell us how fast we are answering, but not what we are answering about. The product team asks "what are customers actually struggling with?" and the support team replies "everything, mostly the usual" because there is no time to comb through hundreds of tickets and look for patterns.

How GloriaMundo Solves It

We build a scheduled workflow that runs once a week. An integration step pulls every ticket resolved in the past seven days from the helpdesk. A code step strips the threads down to the customer's actual question, dropping agent boilerplate and signatures. An LLM step clusters the cleaned tickets into themes — onboarding confusion, billing surprises, specific feature requests, a particular integration failing — and counts how many tickets fall into each. For each theme, a second LLM step picks two or three representative customer quotes that capture the pain in the customer's own words, and writes a one-line summary. The digest is posted to a product Slack channel as a scannable list, and the full version with every quote and ticket link goes into a dated Notion page so anyone can drill in. Glass Box preview shows you the proposed clusters before posting, so you can adjust the theme granularity and the channel.

Example Workflow Steps

  1. Trigger (scheduled): Runs every Monday at 09:00.
  2. Step 1 (integration): Fetch all tickets resolved between last Monday and Sunday from the helpdesk.
  3. Step 2 (code): Clean each ticket thread — strip signatures, footers, and agent boilerplate, keeping the customer's words.
  4. Step 3 (LLM): Cluster the cleaned tickets into themes and assign each ticket a theme label with a confidence score.
  5. Step 4 (LLM): For each theme, pick two or three customer quotes that best represent the issue and write a one-line summary.
  6. Step 5 (integration): Save the full digest — themes, counts, quotes, ticket links — to a dated page in the product Notion workspace.
  7. Step 6 (integration): Post a scannable summary to #product-feedback on Slack with theme names, counts, and a link to the Notion page.

Integrations Used

  • Zendesk — source of the resolved support tickets
  • Notion — stores the full weekly digest for the product wiki
  • Slack — posts the scannable summary to the product team channel

Who This Is For

Heads of support, product managers, and customer experience leads at teams resolving more than fifty tickets a week, where the qualitative signal in the tickets is currently going unread.

Time & Cost Saved

Doing this by hand means a senior support lead spending an afternoon — three to four hours — reading through a week's worth of tickets, taking notes, picking quotes, and writing up the digest. Most teams either do not do it at all or assign it to whoever can spare an afternoon, which means it happens once a quarter rather than once a week. This workflow produces a useful, quote-backed digest every Monday with a review pass of about ten minutes. The workflow uses scheduled, integration, code, and LLM steps, costing a handful of credits per week.